Using the Kurtosis Measure to Identify Clusters in Wireless Channel Impulse Responses

In wireless channel propagation, multipath arrivals appear at the receiver in clusters. Because the notion of clusters tends to be intuitive rather than well-defined, cluster identification in channel impulse responses has traditionally been carried out through human visual inspection. Besides time-consuming for large-scale measurement campaigns, this approach is subjective and will vary from person to person, leading to arbitrary identification. In response to these concerns, automatic clustering algorithms have emerged in the past decade. Most, however, are laden with settings which are sensitive to different radio-frequency environments, again subject to arbitrariness. In this communication we propose a novel clustering algorithm based on the kurtosis measure which, in related work, has been employed precisely for its channel independence. We compare ours to some recent algorithms through a standard validation procedure on simulated impulse responses. We show the proposed algorithm to deliver better results and, because it requires no channel-specific settings, is inherently robust to different environments. Results of the proposed algorithm are also illustrated on real measurements in four buildings with assorted wall materials.